Inside and out: How growers are using AI to improve crop yields and boost business

Photo © Adobestock

Artificial intelligence presents both incredible potential and practical challenges in the green industry. The value of AI as a tool in a grower’s toolbox is still being explored, and that exploration often starts with the simple act of trying to understand what the technology is capable of.

“We just started figuring out what AI was about a year and a half, two years ago,” explains Brett Cromly, general manager of Monrovia’s Oregon nursery. “We knew it was something that was up and coming, and the thought process was: How do we get this to be useful?”

That answer wasn’t immediately apparent. Some creative thinking was required to find an application that made sense, but now, Monrovia uses AI both in the office and in the field.

As more professionals apply the new technology to their operations and their products, a clearer picture of AI’s horticulture capabilities is emerging.

AI takes to the air

Cromly notes the emergence of AI occurred concurrently with the emergence of agricultural drones as useful nursery tools. Of the two technologies, Monrovia adopted drones first, largely to have eyes in the sky to monitor nursery crops. The result was not only a better view of current stock and possible problems, but also a rich set of photographic data. And Cromly says that data was the key to coupling drones with AI.

Around the beginning of 2024, Monrovia’s Oregon location started testing a drone-enabled AI application that would allow the nursery to count its plants. Using the drone imagery, AI has learned how Monrovia spaces its plants and can recognize crop type from lot to lot. With this knowledge, the application can pinpoint each plant and provide accurate cycle counts to make sure inventory numbers reflect reality in the field.

“We’ve been testing it for about a year now. We’re about 99.5% accurate,” Cromly says. “It’s actually proven to be more accurate than if someone went out there hand-clicking.”

A single drone and an AI tool that can provide accurate counts has had immediate real-world savings in labor costs for the company of between $70,000 and $80,000 per year.

Cromly says that Monrovia is currently looking at ways to use AI to interpret the wealth of historical data available in the aerial photographs. He’s currently exploring ways the data can make the nursery more proactive.

“We have limitless data points. How do we use them, and are we using them the right way?” he asks.

AI-enabled humanoid bots?

Bringing AI under cover from the field has the potential to revolutionize the way indoor growers nurture their plants. That revolution is largely dependent on pairing AI with existing automation to further push greenhouse controls toward nearly autonomous operations that can proactively make choices about what’s best for the crop.

According to GrowDirector CEO Dima Chernobilsky, AI technology is already available to provide deep insights into greenhouse operations and plant health, all while making those operations even more efficient. The company recently launched its fourth-generation greenhouse controller and touts AI integration that it claims provides growers with a virtual agronomist.

Chernobilsky says GrowDirector began experimenting with AI in 2024, largely as a “fun project” led by the company's own head agronomist. It was an iterative process, and there were pitfalls in the AI development. “At first, it was lying and doing stupid stuff. It took trial and error to fine-tune it and make it a useful tool instead of a game,” Chernobilsky explains. But he says the integrated AI is now market-ready and requires no additional training by growers once the system is in place.

The GrowDirector system also includes an AI summary with insights and suggestions from greenhouse system usage — AI support that “reads the user manual” so the grower doesn’t have to, plus an AI-enabled CRM called AgriNotes that records any sensor input and provides actionable information.

Chernobilsky predicts that AI will be fully integrated into “physical humanoid robots.” He notes that while it sounds like science fiction, it makes more sense than the current paradigm.

“The environment is built for humans. Nowadays, we need to fine-tune machines to fit the environment or build the environment around them just for one or two functions,” he shares. “A humanoid robot will easily adapt because the world is created for humans.”

That idea fits into Chernobilsky’s philosophy that any AI tool should fit to the user rather than the user having to change to fit the tool. He notes that for those looking for a specific horticultural application, it’s probably on the way. “And even if the tool that they picked is not the right tool, they’re changing so fast, they’re upgrading so fast, that maybe tomorrow, it will be the right tool,” he says.

Patrick Alan Coleman is editor of Greenhouse Management magazine. Contact him at pcoleman@gie.net.